“Learning health care” for patients and populations

In this issue of the Journal, two articles report strikingly high use of health services in Australia.1,2 Lowthian and colleagues document a 75% increase in emergency ambulance transportations, and forecast another 46%–69% increase by 2015.1 Rosenwax and colleagues demonstrate that 96% of Western Australians dying of life-limiting illnesses were admitted to hospital in their last year of life.2 These authors raise three points: use of health care is expanding; solutions are needed to care for people with chronic complex illness; and we can improve understanding and solutions through analysis of growing datasets.

Both studies use linked, large database analyses to understand care patterns. Remarkably, neither proposes algorithms to repeat the analyses at prespecified intervals to monitor change, link findings to other work through data integration, and test interventions.

Rising health care use is obviously unsustainable, and the strain will magnify with more elderly people suffering coexisting chronic illnesses. Individual patient needs are also evolving, and here lies a critical tension. The scaling back of use of health care services seems straightforward; equally straightforward is the decision to admit this woman with refractory cancer pain to hospital, or to call an ambulance for this man with presyncope and facial numbness. Few contemporary patients have simple illnesses; it isn’t “just cancer”, but is a woman with multiple treatments for breast cancer, metastases, osteoporosis, heart disease and emphysema, any of which can stimulate hospitalisation; thus, health care use balloons.

How do we reduce health care use — or at least stabilise it? Health promotion and better prevention is the obvious answer but the processes leading to potentially preventable illnesses among our ageing population are already well underway and unlikely to be substantially modified. A solution is needed, and we must find the right blend of interventions for patients and populations. Data-driven prognostication should guide understanding of an illness trajectory and facilitate decision making. Treatments should be personalised and interventions matched to the patients most likely to benefit. Continuous monitoring of outcomes and adverse effects should be used to ensure promotion of helpful interventions and discontinuation of ineffective or burdensome interventions.

In other words, it’s time to make the transition from isolated findings and single research reports to “learning health care”.3 The care of individuals should be improved by the use of information about all preceding people with similar clinical scenarios, and information about care of a patient reinvested into the growing body of linked data to guide care in the future. In this paradigm, use of available data and evidence will ensure quality of treatments.

The learning health care system forms around three fundamental purposes, which are to:

• generate and apply the best evidence relevant to each patient;

• propel scientific discovery as a “natural outgrowth of patient care”; and

• support quality assessment and improvement, spark innovation, enhance patient safety and maximise health care value.3

According to the United States Institute of Medicine:

Driving the learning health care system is a powerful, integrated, linked data network. Data sources include clinical care, health resource and administrative data, basic science information, clinical research datasets, and patient-reported outcomes.5,6 Patient-level data are used to inform individual care and summarised to improve general health care. Advances in information and data analysis technology will help clinicians temper professional judgement with clinically relevant evidence.

Rapidly increasing health care costs within the context of the need to take care of individual people presents an important contemporary conundrum. A simple solution is not possible. We need a patient-focused learning health system to facilitate best care of individuals and target the right treatment at the right time, while reducing waste and harm. Better care of individuals wraps up to better care of populations and society.